🦀 Bismark 🦀 - my own oxidised rust edition has now reached beta territory and is ready to try out. For the most part it’s a faithful reimplementation: the Rust tools produce byte-identical output to Bismark v0.25.1, checked against the Perl originals as a CI gate (details here). Same results — just the speed, scaling and packaging you’d expect from a Rust rewrite. Confirmed working as a drop-in for nf-core/methylseq — a small nextflow.config profile change points the Bismark steps at the Rust container (byte-identical methylation calls). We ran it end-to-end on a real life experiment on Seqera Platform — selecting the branch + profile

bismark-rust

, nothing else changed (apart from the run time 🙂) I also took the opportunity to add and change a few things, e.g.: New experimental modes — e.g. a combined CT+GA genome index (align against one combined index instead of the usual 2–4), an opt-in memory optimisation for non-directional libraries (one single aligment pass instead of 4 threads!). Preliminary results suggest the combined handling wins for wall time and CPU work for all conditions (some benchmarks) - and scales Note that it is a beta and very much actively developed — plenty is still on the list. One example: rammap, a pure-Rust minimap2 replacement, which would be most relevant for long-read EM-seq (PacBio/ONT) — still tbd. 📖 New docs (and how to install / try it): https://felixkrueger.github.io/Bismark%7Cfelixkrueger.github.io/Bismark (a huge shout-out to my designer-in-chief @Phil Ewels!). Please expect more and better documentation, benchmarks and other improvements soon. Feedback very welcome!